DESeq2 linear contrast design

Question

[kehoe]

I have a design contrast in DESeq2 as follows:

I have three sample types, primordial cells, primary cells, and secondary cells. Cell development ensues from the primordial stage onward. We are interested to study gene expression changes during this development.

Thus, primordial, primary, and secondary groups are categorised as A, B, C, respectively. To elucidate down- and up- regulation of genes going from A to C the DESeq2 contrast would be designed as:

A_to_B <- results(dds, alpha = 0.05, filterFun = ihw, contrast = c("type", "B", "A"))
B_to_C <- results(dds, alpha = 0.05, filterFun = ihw, contrast = c("type", "C", "B"))

Is this correct?

It was written as below previously but I believe this is incorrect. Do you agree?

A_to_B <- results(dds, alpha = 0.05, filterFun = ihw, contrast = c("type", "A", "B"))
B_to_C <- results(dds, alpha = 0.05, filterFun = ihw, contrast = c("type", "B", "C"))

[ATpoint]

The contrasts are identical, only the direction changes.

[kehoe]

Indeed, following what is written in the manual however, shouldn't it be written as the first example?

[ATpoint]

It doesnt matter, it depends on whether you want positive fold changes or negative fold changes when genes in one of the two samples are up- or downregulated.